Method, system and medium for monitoring a laser drilling process based on acoustic emission signals

Abstract

This application discloses a method, system, and medium for monitoring the laser drilling process based on acoustic emission signals, relating to the fields of laser processing and advanced manufacturing. The method includes: performing a short-time Fourier transform on the original audio signal to obtain the spectrum of each frame; extracting the energy of each frame's spectrum within the band of interest to obtain an energy sequence; processing the energy sequence, marking frames exceeding a threshold, and forming threshold intervals based on consecutive threshold frames; merging adjacent threshold intervals with an interval time less than or equal to a preset maximum allowable gap to obtain a merged interval; selecting the merged interval with the longest duration from the merged intervals, determining the start time of the selected merged interval as the processing start time, and the end time as the hole drilling completion time; and determining the time of the last threshold frame in the denoised audio signal as the processing end time. This application achieves accurate monitoring of the processing start, hole drilling completion, and processing end times, improving the automation level and reliability of the monitoring.

Description

Technical Field

[0001] This application relates to the fields of laser processing and advanced manufacturing, and in particular to methods, systems and media for monitoring laser drilling processes based on acoustic emission signals. Background Technology

[0002] During laser drilling, the material is heated and ablated to generate plasma, which is then blown out by an auxiliary gas. This process avoids the problems of delamination and pull-out associated with traditional mechanical drilling, making it particularly suitable for precision hole machining of fiber-reinforced polymer (FRP) composites. However, the thermal effect of laser processing creates a heat-affected zone (HAZ) around the target area. Simultaneously, hole wall reflection and energy attenuation can easily lead to hole taper. Currently, research on reducing the HAZ and taper mainly relies on microscopic observation of the hole wall cross-section after processing. This method damages the workpiece and cannot obtain real-time processing status, resulting in the loss of valuable time-frequency information.

[0003] To monitor the laser drilling process in real time, various methods have been proposed in the industry. Patent CN116532825B collects signal evolution information, hole depth, and motion trajectory to construct a penetration time prediction model and a state recognition model. This method relies on multi-sensor fusion, resulting in complex equipment and failing to fully utilize the time-frequency characteristics of acoustic emission signals. Patent CN119525780A collects signals using an acoustic emission sensor, only determining whether the amplitude exceeds a set threshold to shut down the laser. Patent CN117929355A uses optical methods to monitor the laser-induced breakdown spectrum, but is susceptible to interference from material reflection light and ambient stray light.

[0004] The aforementioned monitoring methods based on acoustic emission sensors only focus on signal amplitude, ignoring the time-frequency information of frequency changes over time. Acoustic emission signals contain rich frequency components during laser drilling, with different processing stages (such as drilling in, drilling through, and hole finishing) corresponding to different time-frequency characteristics. Relying solely on amplitude thresholds cannot accurately identify the drilling through moment and the processing end moment, making it difficult to distinguish between drilling and hole finishing stages, which is detrimental to the fine-grained optimization of process parameters. Therefore, a method is needed to extract the time-frequency characteristics of acoustic emission signals and automatically identify key stages in the drilling process.

[0005] Furthermore, background noise from equipment vibration and airflow occurs at laser drilling sites, and its intensity varies with the environment. Current technologies use fixed thresholds to determine signal amplitude, which cannot adapt to different experimental conditions or different batches of workpieces, easily leading to missed detections or false triggers. Therefore, a method is needed that can dynamically set the detection threshold based on actual environmental noise to improve the robustness of monitoring.

[0006] Furthermore, in laser drilling of composite laminated materials, such as metal-fiber composite laminates, the acoustic emission characteristics of different material layers differ significantly. Existing general time-frequency analysis methods are not optimized for such materials, making it difficult to effectively focus on the characteristic frequency bands of the processed acoustic signals, resulting in insufficient analysis accuracy. Therefore, a time-frequency analysis framework specifically designed for composite laminated materials, capable of pre-setting the frequency band of interest based on material properties, is needed. Summary of the Invention

[0007] The purpose of this application is to provide a method, system, and medium for monitoring the laser drilling process based on acoustic emission signals. This method can extract the time-frequency characteristics of acoustic emission signals, automatically identify the start time of processing, the hole completion time, and the end time of processing, and dynamically set the detection threshold according to the actual environmental noise. At the same time, it can preset the frequency band of interest for composite laminated materials, providing a quantitative basis for the optimization of laser drilling process.

[0008] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a method for monitoring the laser drilling process based on acoustic emission signals, including: Acoustic emission signals generated during laser drilling are collected to obtain the raw audio signal; The original audio signal is denoised to obtain a denoised audio signal. Perform a short-time Fourier transform on the denoised audio signal to obtain the spectrum corresponding to each time frame; A Fourier transform is performed on a segment of continuous signal in the initial part of the original audio signal to obtain the noise spectrum, and an energy threshold is set according to the energy of the noise spectrum within a preset frequency band of interest. Based on the frequency band of interest, the energy of the spectrum corresponding to each time frame within the frequency band of interest is extracted to obtain an energy sequence; The energy sequence is smoothed to obtain a smoothed energy sequence; All frames in the smoothed energy sequence that are greater than the energy threshold are marked as over-threshold frames, and over-threshold intervals are formed based on consecutive over-threshold frames. Merge adjacent over-threshold intervals with an interval time less than or equal to the preset maximum allowable gap to obtain a merged interval; Select the merged interval with the longest duration from all merged intervals, and determine the start time of the selected merged interval as the processing start time, and the end time of the selected merged interval as the hole drilling completion time. The moment of the last frame exceeding the threshold in the noise-reduced audio signal is determined as the processing end moment; Output the processing start time, the hole drilling completion time, and the processing end time.

[0009] Optionally, noise reduction is performed on the original audio signal, specifically including: The noise signal was obtained by collecting pure ambient noise during laser-free processing. Perform a short-time Fourier transform on the noise signal to obtain the amplitude spectrum of the noise signal in each time frame; The average noise amplitude spectrum is obtained by averaging the amplitude spectra of each time frame. Perform a short-time Fourier transform on the original audio signal to obtain the amplitude spectrum and phase spectrum of the original audio signal at each time frame; Based on the amplitude spectrum of the original audio signal at each time frame and the average noise amplitude spectrum, the enhanced amplitude spectrum is obtained by spectral subtraction. Based on the enhanced amplitude spectrum and the phase spectrum of the original audio signal at each time frame, the time-domain signal is reconstructed to obtain the denoised audio signal.

[0010] Alternatively, the enhanced amplitude spectrum can be calculated using the following formula: ; in, The amplitude spectrum value of the original audio signal at the m-th time frame and the k-th frequency index is... The value of the average noise amplitude spectrum at the k-th frequency index. For over-subtraction factor, It is 1.5; The lower limit factor of the spectrum. It is 0.01; For time frame indexing, For frequency indexing.

[0011] Optionally, the frequency band of interest is predetermined based on the modal frequency or experimental power spectrum of the workpiece to be processed, with a lower limit of 2000 Hz and an upper limit of 8000 Hz.

[0012] Optionally, the method for setting the energy threshold is as follows: take the energy values ​​of a preset number of consecutive time frames in the starting part of the original audio signal within the frequency band of interest, calculate the median of the energy values ​​as the noise floor, and multiply the noise floor by a preset threshold multiple to obtain the energy threshold.

[0013] Alternatively, the following formula can be used for smoothing: ; in, The energy within the band of interest for the time frame before the b-th smoothing process. To smooth the window length, It is an odd number; For time frame indexing, For the smoothed first The energy of the spectrum corresponding to each time frame within the band of interest.

[0014] Optionally, after outputting the processing start time, the hole drilling completion time, and the processing end time, the total processing time, drilling time, and hole finishing time are calculated according to the following formula: ; ; ; in, The start time of processing. When the hole is drilled through, The end time of processing Total processing time For drilling time, This refers to the time required for hole repair.

[0015] Secondly, this application provides a laser drilling process monitoring system based on acoustic emission signals, comprising: Lasers are used for laser drilling of workpieces. Acoustic emission sensors are used to convert acoustic signals generated during drilling into electrical signals; A data acquisition unit, connected to the acoustic emission sensor, is used to acquire the electrical signal and perform analog-to-digital conversion to obtain the original audio signal; A processor, connected to the data acquisition unit, is configured to execute the aforementioned laser drilling process monitoring method based on acoustic emission signals.

[0016] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described laser drilling process monitoring method based on acoustic emission signals.

[0017] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a method, system, and medium for monitoring laser drilling processes based on acoustic emission signals. The method involves denoising the acquired raw audio signal to obtain a denoised audio signal, and then performing a short-time Fourier transform on the denoised audio signal to extract the energy sequence within a preset frequency band of interest. Through smoothing, over-threshold frame marking, and interval merging, the method automatically identifies the start time of processing, the hole completion time, and the processing end time. This fully utilizes time-frequency information and overcomes the shortcomings of existing technologies that rely solely on amplitude thresholds, resulting in the loss of time-frequency information and the inability to distinguish between drilling and hole finishing stages. Simultaneously, this application utilizes the continuous signal at the beginning of the raw audio signal to perform a Fourier transform to obtain a noise spectrum, and dynamically sets an energy threshold based on the energy within the frequency band of interest according to this noise spectrum. This solves the problem of missed detections or false triggers caused by fixed thresholds and improves environmental adaptability. Furthermore, this application focuses energy extraction on the frequency range where the processing acoustic signal is concentrated by preseting the frequency band of interest, providing a dedicated time-frequency analysis framework for the laser drilling process of composite materials, improving monitoring accuracy and process optimization capabilities. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the laser drilling process monitoring method based on acoustic emission signals according to this application. Figure 2 A schematic diagram illustrating the power spectral density of the band of interest; Figure 3 This is a schematic diagram illustrating the extraction of processing time using the method described in steps 101 to 111. Figure 4 This is a 3D time-frequency-amplitude diagram of the method described in steps 101 to 111; Figure 5 This is a 3D time-frequency-amplitude diagram using the short-time RMS method; Figure 6 This is a 3D time-frequency-amplitude diagram using the short-time RMS method. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] To make the objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is applicable to metal / fiber-reinforced polymer (FRP) composite laminates and is a method for determining drilling status and identifying processing stages based on time-frequency domain analysis of acoustic emission signals during processing.

[0022] The specific parameters used in this embodiment are shown in Table 1.

[0023] Table 1

[0024] The apparatus used to implement the method of this application includes: a laser, an acoustic emission sensor, a processing table, a fixture, a data acquisition device, and a terminal computer.

[0025] The laser is used to emit a laser beam to perform laser drilling on the workpiece and generate an acoustic signal source; in this embodiment, an ultraviolet nanosecond pulse laser with a repetition frequency of 50kHz is used.

[0026] The acoustic emission sensor is fixedly installed near the workpiece to be processed, for example, by means of a magnetic base or adhesive, and is used to convert the acoustic signal generated during drilling into an electrical signal.

[0027] The data acquisition unit is connected to the acoustic emission sensor to acquire the electrical signal and perform analog-to-digital conversion to obtain the original audio signal; in this embodiment, the sampling rate is set to 44.1kHz.

[0028] The terminal computer is connected to the data acquisition unit and is equipped with data processing software and its signal processing toolbox, such as MATLAB and Signal Processing Toolbox and DSP System Toolbox, or Python (ANACONDA distribution) and WaLSAtools toolkit, etc., to analyze the acoustic characteristics of the laser processing process converted into digital signals and determine the real-time processing status.

[0029] Before processing, initialize the acoustic emission sensor and data acquisition unit, and check the connection and driver programs between the terminal computer and the data acquisition unit. Write scripts on the terminal computer for Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), spectral subtraction noise reduction, and visualization.

[0030] Furthermore, the acoustic information sampling steps during processing are as follows: Before processing begins, a segment of pure noise is recorded to simulate interference from the experimental environment. During processing, the acoustic emission sensor remains enabled, and a high-sampling-rate audio file is recorded. After processing, the audio file is analyzed on the terminal computer.

[0031] In this embodiment, based on the above-described device, such as Figure 1 As shown, a method for monitoring the laser drilling process based on acoustic emission signals is provided. During laser drilling, the material is heated and impacted, generating strong acoustic emission, resulting in a signal amplitude significantly higher than the background noise. When the hole is drilled through but the material has not completely fallen out, or when the laser can still interact with the taper, the time-frequency domain energy changes significantly, but remains significantly higher than the background. After processing, the acoustic signal returns to the background level. Therefore, by monitoring the short-term energy changes in the signal amplitude, the start, drilling completion, and end times of processing can be located. Specifically, it includes the following steps 101 to 111. Wherein: Step 101: Collect the acoustic emission signal generated during laser drilling to obtain the original audio signal.

[0032] Before processing begins, a 0.5-second segment of pure ambient noise (noise without laser processing) is recorded to simulate interference from the experimental environment and serve as the noise signal. Laser drilling is then initiated, and the acoustic emission sensor is activated simultaneously to record the audio signal of the entire drilling process. After recording, the audio file is imported into the terminal computer. If the original audio is stereo, the average of the corresponding sampling points in the left and right channels is taken to convert it to mono; the sampling rate of 44.1kHz is kept constant, and no downsampling is performed to preserve complete time-frequency information. This yields the original audio signal.

[0033] Step 102: Denoise the original audio signal to obtain a denoised audio signal.

[0034] Step 103: Perform a short-time Fourier transform on the denoised audio signal to obtain the spectrum corresponding to each time frame.

[0035] The parameters of the short-time Fourier transform are as follows: the window function is a Hamming window, and the window length is... The frame has 1024 points, corresponding to a time length of approximately 23.2 milliseconds. The frame shift R is 256 points, corresponding to an overlap rate of 75%, a time resolution of approximately 5.8 milliseconds, and a Fast Fourier Transform (FFT) point count N of 1024. This yields the spectrum X(m,k) for each time frame, where m is the frame index (m=1,2,…,M, M is the total number of frames) and k is the frequency index (k=0,1,…,1023). The frequency resolution Δf = 44100 / 1024 ≈ 43.07 Hz. The Short-Time Fourier Transform is: k=0,1,…,N 1; Where w[n] is the value of the window function at the nth point, x[n+mR] is the value of the nth sampling point in the mth frame, and X[m,k] is the spectral value of the kth frequency point in the mth frame.

[0036] Step 104: Perform a Fourier transform on a segment of continuous signal in the initial part of the original audio signal to obtain the noise spectrum, and set an energy threshold based on the energy of the noise spectrum within a preset frequency band of interest.

[0037] In one preferred embodiment, the frequency band of interest is predetermined based on the modal frequencies or experimental power spectrum of the workpiece to be processed, such as... Figure 2 As shown, the lower limit of the frequency band of interest is 2000 Hz, and the upper limit is 8000 Hz. This setting focuses energy extraction on the frequency band where the processing sound signal is concentrated, effectively filtering out low-frequency mechanical vibrations (<500Hz) and high-frequency environmental noise (>10kHz), thus improving the signal-to-noise ratio.

[0038] As another preferred embodiment, the method for setting the energy threshold is as follows: take the energy values ​​of a preset number of consecutive time frames in the starting part of the original audio signal within the frequency band of interest, calculate the median of the energy values ​​as the noise floor, and multiply the noise floor by a preset threshold multiple to obtain the energy threshold.

[0039] Specifically, take the beginning part of the original audio signal, i.e., the signal before noise reduction. frame, =10, corresponding to the pure noise segment before processing begins. Each frame signal undergoes a Fourier transform with the same parameters as in step 103 to obtain the spectrum of each frame. The energy of these spectra within a preset band of interest is then calculated. This yields the previous... After considering all energy values, the median is taken as the noise floor. : ; in, The energy of the spectrum corresponding to the first time frame within the frequency band of interest. The energy of the spectrum corresponding to the second time frame within the frequency band of interest. For the first The energy of the spectrum corresponding to each time frame within the band of interest.

[0040] Take the threshold multiple λ=5, which can be adjusted according to the signal-to-noise ratio. Calculate the energy threshold. : This adaptive threshold setting allows the energy threshold to automatically adjust according to the actual environmental noise level, eliminating the need for repeated manual adjustments. Using the median instead of the mean effectively avoids the impact of individual abnormal frames on background estimation.

[0041] Step 105: Based on the frequency band of interest, extract the energy of the spectrum corresponding to each time frame within the frequency band of interest to obtain an energy sequence.

[0042] Specifically, for the spectrum X(m,k) of each time frame obtained in step 103, its energy within the band of interest is calculated using the following formula: ; in, For the first The energy of the spectrum corresponding to each time frame within the frequency band of interest. This is the set of frequency indices corresponding to the frequency band.

[0043] Step 106: Smooth the energy sequence to obtain a smoothed energy sequence.

[0044] Specifically, to suppress impulse interference, a moving average is applied to the energy sequence: ; in, The energy within the band of interest for the time frame before the b-th smoothing process. To smooth the window length, It is an odd number; For time frame indexing, For the smoothed first The energy of the spectrum corresponding to each time frame within the frequency band of interest. The moving average suppresses the energy drop-off between individual pulses in pulsed laser processing, enabling the over-threshold range to continuously cover the entire processing period and avoiding interval breaks caused by brief energy troughs.

[0045] Step 107: Mark all frames in the smoothed energy sequence that are greater than the energy threshold as over-threshold frames, and form an over-threshold interval based on consecutive over-threshold frames.

[0046] Specifically, For each frame m, if If a frame is found to be above the threshold, it is marked as a frame with B(m) = 1; otherwise, it is a frame not above the threshold, and B(m) = 0. Traverse all frames and find all consecutive sequences of frames marked as 1. Each such consecutive sequence constitutes a threshold interval. Each threshold interval is defined by its starting frame index and ending frame index.

[0047] Step 108: Merge adjacent over-threshold intervals with an interval time less than or equal to the preset maximum allowable gap to obtain a merged interval.

[0048] Specifically, identify all consecutive intervals exceeding the threshold. If the time interval between adjacent intervals... If so, then they will be merged into the same interval. The maximum allowable gap is used to cope with brief energy drops during the machining process, such as pulse intervals, to prevent the pulse signal trough in the later stages of drilling from being mistaken for the end of machining or confused with drilling through.

[0049] The center time of each frame is calculated from the frame index: frame shift 256 points, window length 1024 points, half-window length 512 points, center time t(m) = [(m) [1)×256+512] / 44100 seconds. Let the center time of the ending frame of the preceding interval be [1)×256+512] / 44100 seconds. The center time of the starting frame of the next interval is Then the interval time ,like ,in, =0.5 seconds, which merges the two intervals into one interval. Repeat this process until there are no more adjacent intervals to merge, resulting in one or more merged intervals.

[0050] Step 109: Select the merged interval with the longest duration from all merged intervals, and determine the start time of the selected merged interval as the processing start time, and the end time of the selected merged interval as the hole drilling completion time.

[0051] Calculate each merge interval The duration is calculated as the difference between the end frame center time and the start frame center time: The longest merging interval is selected as the main processing segment. The center time of the starting frame of this merging interval is recorded as the processing start time. The center time of the end frame of the merged interval is recorded as the hole drilling completion time. .

[0052] Step 110: Determine the time of the last frame exceeding the threshold in the denoised audio signal as the processing end time. In the entire denoised audio signal, find the last frame exceeding the threshold, i.e., the largest m such that B(m) = 1, and record the center time of this frame as the processing end time. .

[0053] Step 111: Output the processing start time, the hole drilling completion time, and the processing end time, as follows: Figure 3 As shown, the red dashed line represents the start time of machining at 12.88 s, the green dashed line represents the hole drilling completion time at 82.70 s, and the blue dashed line represents the end time of machining at 94.77 s. In a preferred embodiment, the machining start time is output... Hole drilling time and processing end time Then, calculate the total processing time, drilling time, and hole-fixing time according to the following formula: Total processing time: ; Drilling time ; Repair time ; These quantitative indicators automatically identify the key moments of the start, drilling, and end of the process through audio signals, directly quantifying the drilling time, hole repair time, and total processing time. They can reflect the processing efficiency and process characteristics under different process parameters in real time and non-destructively, thus providing immediate and continuous data for parameter optimization and avoiding the drawbacks of traditional hole cutting inspection that require workpiece destruction and are cumbersome.

[0054] By implementing steps 101 to 111 above, this application denoises the acquired original audio signal, performs a short-time Fourier transform on the denoised audio signal to extract the energy sequence within a preset frequency band of interest, and merges adjacent over-threshold intervals with an interval less than or equal to the preset maximum allowable gap through smoothing, over-threshold frame marking, and merging. The longest duration is selected from all merged intervals, and its start and end times are determined as the processing start time and hole drilling completion time, respectively. The time of the last over-threshold frame in the denoised audio signal is determined as the processing end time. This fully utilizes time-frequency information to overcome the shortcomings of existing technologies that rely solely on amplitude thresholds, resulting in the loss of time-frequency information and the inability to distinguish between drilling and hole finishing stages. Simultaneously, this application uses the continuous signal at the beginning of the original audio signal to perform a Fourier transform to obtain the noise spectrum, and dynamically sets the energy threshold based on the energy of this noise spectrum within the preset frequency band of interest, solving the problem of poor adaptability of fixed thresholds. Furthermore, by focusing energy extraction on the frequency range where the processing acoustic signal is concentrated through the preset frequency band of interest, a dedicated time-frequency analysis framework is provided for the laser drilling process of composite laminated materials, achieving refined monitoring of the drilling process and quantitative optimization of process parameters.

[0055] Furthermore, noise reduction is performed on the original audio signal, specifically including: Collect the pure environmental noise without laser processing to obtain a noise signal; read the noise signal and convert it to mono; if the sampling rate is inconsistent with the original audio signal, use the resampling function to resample it to the same sampling rate , to ensure frequency domain matching.

[0056] Perform a short-time Fourier transform on the noise signal to obtain the amplitude spectrum of the noise signal on each time frame; Before processing the formal processing audio file, perform a fast Fourier transform and a short-time Fourier transform on the noise signal alone to obtain its time-frequency characteristics.

[0057] Among them, the fast Fourier transform is used to obtain a frequency-amplitude spectrum. For a discrete sequence of length , where , , is the number of points of the fast Fourier transform, and its discrete Fourier transform (Discrete Fourier Transform, DFT) is defined as:

[0058] where x[n] is the nth sampling point of the input, k is the frequency index, X[k] is the spectrum value of the kth frequency point, corresponding to the frequency , is the sampling rate.

[0059] The short-time Fourier transform is used to obtain a time-frequency spectrum. For the discrete signal x[n], the sampling rate is , let the window length be L, corresponding to the time length L / fs, the window function be w[n], and the frame shift be R. Usually, R < L and R < L, because there is overlap between frames, its discrete short-time Fourier transform is defined as: ; where w[n] is the value of the window function at the nth point, x[n + mR] is the nth sampling point value in the mth frame, and X[m, k] is the spectrum value of the kth frequency point in the mth frame.

[0060] According to the frequency-amplitude spectrum and time-frequency spectrum obtained above, a 3D time-frequency-amplitude spectrum is obtained. The 3D time-frequency-amplitude analysis diagram can clearly show the change rate and degree of the signal in each analyzed dimension, and can be rotated and enlarged to a certain extent to further observe details, such as Figure 4 shown, which helps to deeply analyze the time-frequency characteristics of the acoustic emission signal during the drilling process.

[0061] The average noise amplitude spectrum is obtained by averaging the amplitude spectra across all time frames. A short-time Fourier transform (SFT) is performed on the noise signal using the same parameters as those used in the subsequent formal audio processing SFT analysis. The absolute value of the amplitude spectrum for each frame is taken, and then averaged along the time axis (frame direction) to obtain the average noise amplitude spectrum. (Column vector, length N / 2+1). This averaging operation utilizes the assumption of noise stationarity, making the estimate more stable.

[0062] Perform a short-time Fourier transform on the original audio signal to obtain the amplitude spectrum of the original audio signal at each time frame. and phase spectrum ; Based on the amplitude spectrum of the original audio signal at each time frame and the average noise amplitude spectrum, the enhanced amplitude spectrum is obtained by spectral subtraction; for each time frame index m and frequency index k, the enhanced amplitude spectrum is calculated using the following formula: ; in, For time frame indexing, For frequency index, The amplitude spectrum value of the original audio signal at the m-th time frame and the k-th frequency index is... The value of the average noise amplitude spectrum at the k-th frequency index. For over-subtraction factor, The value is 1.5, used to control the noise reduction intensity; The lower limit factor of the spectrum. The value is 0.01, which is used to retain a certain amount of background noise and avoid generating "musical noise". These two parameters were adjusted through pre-experimentation to balance noise reduction effect and signal distortion.

[0063] Based on the enhanced amplitude spectrum and the phase spectrum of the original audio signal at each time frame, the time-domain signal is reconstructed to obtain the denoised audio signal.

[0064] Specifically, the enhanced amplitude spectrum is combined with the original phase to obtain the enhanced complex spectrum: ; Since the complex spectrum obtained by the short-time Fourier transform is a one-sided spectrum, meaning it only contains the positive frequency portion, it needs to be extended into a complete conjugate symmetric spectrum to satisfy the real signal condition for the inverse transform. The enhanced complex spectrum is then reconstructed into a time-domain signal using the inverse short-time Fourier transform. The inverse short-time Fourier transform uses the same window function and overlap settings as the short-time Fourier transform, and synthesizes the signal through an overlap-addition method. Since numerical computation may introduce a small imaginary part, the real part is taken to obtain the final denoised signal.

[0065] In another implementation, this application can also use the root mean square (RMS) method to extract the processing time, as it is sensitive to transient changes and computationally simple. The specific steps are as follows: Step 201: Convert the noise-reduced audio signal into a single channel with a sampling rate of [missing information]. =44.1kHz. Framing is performed according to the following parameters: frame length L = 1024 points, frame shift R = 256 points, corresponding to an overlap rate of 75%. The total number of frames M is calculated as follows: ; where length(x) is the total number of signal sampling points.

[0066] For each frame , =1,2,…,M, calculate the RMS value: ;in, is the sampling point of the m-th frame.

[0067] Simultaneously calculate the center time of each frame (for time positioning): ; Step 202: Assuming there is pure noise before recording or processing begins, take the median of the RMS values ​​of the first 10 frames (or all frames if the total number of frames is less than 10) as the background noise level. : ; Using the median instead of the mean can effectively avoid the impact of individual abnormal frames (such as sudden interference) on background estimation.

[0068] Step 203: Set the detection threshold A fixed multiple of the background noise level: Among them, multiples This is an empirical value and can be adjusted based on the actual signal-to-noise ratio. This factor ensures that the threshold is higher than the background fluctuations while capturing the energy increase caused by processing.

[0069] Step 204: Iterate through the RMS values ​​of all frames and find all frames that satisfy the condition. Frame index set .

[0070] like If not empty, then the processing start frame index is The end frame index is .

[0071] Corresponding processing start time Processing end time .

[0072] Processing time .

[0073] like If empty, a warning will be output, indicating that the processing period could not be detected automatically (possibly due to an excessively high threshold or a weak signal).

[0074] Step 205: Due to the frame center time The midpoint of the frame is defined, so the detected start and end times correspond to the center of the first and last frames where the processing energy exceeds the threshold. This method can accommodate brief energy drops during processing, such as laser pulse intervals. As long as most frames exceed the threshold during the entire processing period, the start and end times are determined by the first and last frames that exceed the threshold.

[0075] Experiments show that when using the short-time RMS method to extract machining time, under certain process parameter conditions, the machining signal is pulsed, which can easily lead to the pulse signal trough being mistaken for the end of machining or confused with drilling completion. Figure 5 and Figure 6 As shown.

[0076] This application provides two methods for extracting processing time: the short-time RMS method and the method described in steps 101 to 111. The short-time RMS method uses the root mean square in the time domain as the energy metric, is simple to calculate, and is suitable for scenarios with high signal-to-noise ratio and continuous processing signals. The method described in steps 101 to 111 analyzes the processing energy through short-time Fourier transform. It can be used after at least one acoustic analysis has been performed to roughly grasp the frequency domain characteristics of the processing energy, allowing the determination of the start and end times to be concentrated within a certain frequency range, thereby improving the signal-to-noise ratio, enhancing robustness, and enabling more detailed processing status judgments.

[0077] In summary, in practical applications, the methods described in steps 101 to 111 are preferred for obtaining more precise measurements of the three key moments: the start of machining, the hole completion time, and the end of machining. The short-time RMS method can serve as a simplified alternative for rapid evaluation, but it should be used with caution in pulsed laser processing, and verification in conjunction with time-frequency diagrams is recommended.

[0078] The beneficial effects of this application are as follows: By performing a short-time Fourier transform on the acoustic emission signal and extracting the energy within the frequency band of interest, combined with smoothing processing, merging of over-threshold intervals, and selection of the longest interval, this application can automatically output three key moments: processing start, hole drilling completion, and processing completion. This overcomes the shortcomings of existing technologies that rely solely on amplitude thresholds, resulting in the loss of time-frequency information and the inability to distinguish between drilling completion and hole finishing stages. Experiments show that in laser drilling of copper / resin laminates, the method of this application can accurately identify the entire processing process, with a theoretical detection error of less than ±0.02 seconds, significantly outperforming detection methods based on amplitude thresholds.

[0079] This application combines Fast Fourier Transform and Short-Time Fourier Transform to construct a three-dimensional image of time-frequency-amplitude, which can intuitively present the entire process of excitation, maintenance and decay of resonant frequency during processing, overcoming the deficiency that single spectrum analysis cannot reveal time-varying characteristics.

[0080] This application employs spectral subtraction for noise reduction before analysis, estimating the amplitude spectrum by recording pure environmental noise and adaptively subtracting the noise component. In harsh environments with a signal-to-noise ratio (SNR) below 10 dB, the SNR of the denoised signal is improved by an average of 12 to 15 dB, enabling clear identification of weak processed signals. The optimal combination of an over-subtraction factor of 1.5 and a spectral lower limit factor of 0.01 effectively suppresses noise while avoiding the introduction of musical noise.

[0081] This application employs adaptive noise basis estimation, taking the median energy of a preset number of time frames in the initial part of the original audio signal within the band of interest as the noise basis, multiplying it by a threshold factor to obtain the energy threshold. This approach can adapt to background variations in different experimental environments and recording equipment. Key parameters such as the band of interest, threshold factor, smoothing window length, and maximum allowable gap can be flexibly adjusted according to material properties and process characteristics. It is applicable to various material systems such as copper / resin laminates, carbon fiber reinforced composites, and metal single-layer plates, as well as various processing methods such as nanosecond, picosecond, and continuous laser processing.

[0082] The three key moments output in this application, along with the calculated drilling time, finishing time, and total processing time, provide direct quantitative indicators for optimizing laser drilling processes. By analyzing the drilling time under different laser powers, the optimal power range can be quickly determined; by combining the changes in finishing time with microscopic imaging, the heat-affected zone and taper control effects can be evaluated, avoiding the cumbersome and material-wasting nature of traditional destructive hole-cutting inspections.

[0083] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0084] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0085] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for monitoring laser drilling processes based on acoustic emission signals, characterized in that, The method includes: Acoustic emission signals generated during laser drilling are collected to obtain the raw audio signal; The original audio signal is denoised to obtain a denoised audio signal. Perform a short-time Fourier transform on the denoised audio signal to obtain the spectrum corresponding to each time frame; A Fourier transform is performed on a segment of continuous signal in the initial part of the original audio signal to obtain the noise spectrum, and an energy threshold is set according to the energy of the noise spectrum within a preset frequency band of interest. Based on the frequency band of interest, the energy of the spectrum corresponding to each time frame within the frequency band of interest is extracted to obtain an energy sequence; The energy sequence is smoothed to obtain a smoothed energy sequence; All frames in the smoothed energy sequence that are greater than the energy threshold are marked as over-threshold frames, and over-threshold intervals are formed based on consecutive over-threshold frames. Merge adjacent over-threshold intervals with an interval time less than or equal to the preset maximum allowable gap to obtain a merged interval; Select the merged interval with the longest duration from all merged intervals, and determine the start time of the selected merged interval as the processing start time, and the end time of the selected merged interval as the hole drilling completion time. The moment of the last frame exceeding the threshold in the noise-reduced audio signal is determined as the processing end moment; Output the processing start time, the hole drilling completion time, and the processing end time.

2. The laser drilling process monitoring method based on acoustic emission signals according to claim 1, characterized in that, Noise reduction of the original audio signal includes: The noise signal was obtained by collecting pure ambient noise during laser-free processing. Perform a short-time Fourier transform on the noise signal to obtain the amplitude spectrum of the noise signal in each time frame; The average noise amplitude spectrum is obtained by averaging the amplitude spectra of each time frame. Perform a short-time Fourier transform on the original audio signal to obtain the amplitude spectrum and phase spectrum of the original audio signal at each time frame; Based on the amplitude spectrum of the original audio signal at each time frame and the average noise amplitude spectrum, the enhanced amplitude spectrum is obtained by spectral subtraction. Based on the enhanced amplitude spectrum and the phase spectrum of the original audio signal at each time frame, the time-domain signal is reconstructed to obtain the denoised audio signal.

3. The laser drilling process monitoring method based on acoustic emission signals according to claim 2, characterized in that, The enhanced amplitude spectrum is calculated using the following formula: ； in, The amplitude spectrum value of the original audio signal at the m-th time frame and the k-th frequency index is... The value of the average noise amplitude spectrum at the k-th frequency index. For over-subtraction factor, It is 1.5; The lower limit factor of the spectrum. It is 0.01; For time frame indexing, For frequency indexing.

4. The laser drilling process monitoring method based on acoustic emission signals according to claim 1, characterized in that, The frequency band of interest is predetermined based on the modal frequency or experimental power spectrum of the workpiece to be processed. The lower limit of the frequency band of interest is 2000 Hz and the upper limit is 8000 Hz.

5. The laser drilling process monitoring method based on acoustic emission signals according to claim 1, characterized in that, The method for setting the energy threshold is as follows: take the energy values ​​of a preset number of consecutive time frames in the starting part of the original audio signal within the frequency band of interest, and calculate the median of the energy values ​​as the noise floor; The energy threshold is obtained by multiplying the noise floor by a preset threshold factor.

6. The laser drilling process monitoring method based on acoustic emission signals according to claim 1, characterized in that, The following formula is used for smoothing: ； in, The energy within the band of interest for the time frame before the b-th smoothing process. To smooth the window length, It is an odd number; For time frame indexing, For the smoothed first The energy of the spectrum corresponding to each time frame within the band of interest.

7. The laser drilling process monitoring method based on acoustic emission signals according to claim 1, characterized in that, After outputting the processing start time, the hole drilling completion time, and the processing end time, the total processing time, drilling time, and hole finishing time are calculated according to the following formula: ； ； ； in, The start time of processing. When the hole is drilled through, The end time of processing. Total processing time For drilling time, This refers to the time required for hole repair.

8. A laser drilling process monitoring system based on acoustic emission signals, characterized in that, The system includes: Lasers are used for laser drilling of workpieces. Acoustic emission sensors are used to convert acoustic signals generated during drilling into electrical signals; A data acquisition unit, connected to the acoustic emission sensor, is used to acquire the electrical signal and perform analog-to-digital conversion to obtain the original audio signal; A processor, connected to the data acquisition unit, is configured to perform the laser drilling process monitoring method based on acoustic emission signals as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the laser drilling process monitoring method based on acoustic emission signals as described in any one of claims 1-7.

Images